•  
  •  
 

ORCID

Anna Gaweł: 0000-0001-6753-6502

Janusz Kudła: 0000-0003-2485-6877

Keywords

GARCH-MIDAS, asymmetric volatility models, volatility prediction, high-frequency data, COVID-19 pandemic

Abstract

Purpose

We address the problem of forecasting USD/CHF volatility at the beginning of the COVID-19 crisis. We chose popular currencies (Swiss franc and American dollar) in the period 1.07.2020 to 31.12.2020.

Design/methodology/approach

We employed several volatility models, including APARCH, EGARCH, GJR-GARCH, TGARCH, and GARCH-MIDAS, on high-frequency USD/CHF data. Particular emphasis was placed on asymmetric models to capture volatility asymmetry.

Findings

The highest volatility occurred during the first wave of the COVID-19 pandemic. Volatility forecasts are most accurate with EGARCH and GARCH-MIDAS models that incorporate long-term asymmetry, particularly when predicting volatility over a longer planning horizon. GARCH-MIDAS models with short-term asymmetry perform best in the sample but are inferior in forecasting future volatility (out-of-sample).

Originality

The originality refers to the subject of study (exchange rates instead of stocks), the methods used (GARCH-MIDAS, asymmetric volatility models), and the particular crisis period (the outbreak of the COVID-19 pandemic).

Research limitations/implications

Even in a market of relatively low volatility, such as forex, volatility reveals both long- and short-run components during the pandemic crisis and some asymmetry. Therefore, the use of more complicated methods is sometimes not justified by the improvement of prediction accuracy. The results are limited to specific data and a crisis period. Therefore, in the future, we need to determine whether these methods are effective in periods with average volatility.

Acknowledgments

Funding

The research received no funds.

Declaration of Conflicting Interests

The author declared no potential conflicts of interest with respect to the research, authorship, and publication of the article.

Declaration about the scope of AI utilization

The authors did not use an AI tool in the preparation of the article.

First Page

42

Last Page

59

Page Count

18

Received Date

10.10.2024

Revised Date

22.05.2025

Accept Date

21.07.2025

Online Available Date

29.08.2025

DOI

10.7172/2353-6845.jbfe.2025.2.3

JEL Code

C53, F31, G17

Publisher

University of Warsaw

Share

COinS